AN UNMANNED AERIAL SYSTEM FOR ON-THE-EDGE TRAFFIC MONITORING IN WORK-ZONE APPROACH AREA

Abstract

Maintenance works are crucial to improve the reliability and resilience of road infrastructure but, despite efforts to achieve a safer operation, work zones are still risky areas where 4% of all road accidents occur. The main factors that increase the risk during maintenance include the proximity to live traffic, inadequate warning signs and driver behaviour. Intelligent Transportation Systems and their supporting technologies including sensors, data processing and analysis have been beneficial for increasing road safety. In this work, we present the design of a context awareness approach based on an Unmanned Aerial System aimed to detect inadequate speed of incoming traffic approaching to a work zone and to raise warning alerts. To accomplish this objective, an optical payload carrying an on-the-edge analysis system based on deep learning tracking was developed and tested. Preliminary results show the potential of the design to achieve near real-time operation preserving a mean Average Precision similar to that obtained with more complex architectures.